Exploring the role of software architecture in dynamic and fault tolerant pervasive systems l.jpg
1 / 25

Exploring the Role of Software Architecture in Dynamic and Fault Tolerant Pervasive Systems.


55 views
Uploaded on:
Category: Animals / Pets
Description
Exploring the Role of Software Architecture in Dynamic and Fault Tolerant Pervasive Systems. Authors Chiyoung Seo1, Sam Malek1, George Edwards1, Daniel Popescu1, Nenad Medvidovic1, Brad Petrus2, and Sharmila Ravula3. Presenter Edward Gerhardstein. Introduction.
Transcripts
Slide 1

Investigating the Role of Software Architecture in Dynamic and Fault Tolerant Pervasive Systems Authors Chiyoung Seo1, Sam Malek1, George Edwards1, Daniel Popescu1, Nenad Medvidovic1, Brad Petrus2, and Sharmila Ravula3 Presenter Edward Gerhardstein

Slide 2

Introduction The previous couple of decades have seen a development in size and multifaceted nature of programming frameworks. A rising class of installed and pervasive programming framework are developing in notoriety. Because of rate, cost, remote systems, and handheld gadgets. Programming designs give reflections to speaking to the structure, conduct, and key properties of a product framework.

Slide 3

Introduction They are portrayed regarding programming parts (computational components), connectors (association components), And their arrangements. Programming building styles further refine the vocabulary of segment and connector sorts and propose an arrangement of limitations on how occurrences of those sorts might be consolidated in a framework. (e.g., distribute subscribe, shared, pipe-and-channel, customer server)

Slide 4

Introduction For programming design models to be really valuable in an advancement setting, they should be joined by backing for their usage. With implanted and pervasive situations: The creators contend that a structural middleware stage, which gives local execution level backing to the key design reflections, is more qualified than conventional middleware stages to address the product building challenges innate in creating pervasive frameworks.

Slide 5

Introduction Three key offices are required for the advancement of flaw tolerant pervasive frameworks: (1) dynamic disclosure of new administrations and assets, (2) computerized and straightforward recuperation from disappointment, and (3) systematic determination of part replication methodologies and arrangement structures. We demonstrate that by making adaptation to internal failure concerns unequivocal in the structural models, we can investigate and enhance the framework's versatility to startling disappointments. We show that an engineering middleware can give propelled adaptation to internal failure offices, while accomplishing proficiency and structural clarity.

Slide 6

Introduction This idea will be shown through a group of issue tolerant sensor net works. Area 2 gives a diagram of a group of sensor system applications. Area 3 displays a middleware stage that actualizes a design based answer for adaptation to internal failure in pervasive frameworks. Area 4 examines our backing for administration disclosure. Area 5 shows the offices for recuperating from disappointment. Segment 6 introduces an instrument suite for investigating and dissecting a framework's QoS properties, including its adaptation to non-critical failure. Segment 7 introduces the related work. At long last, the paper finishes up with a brief discourse and diagram of future work.

Slide 7

Application Scenario Figure 1. MIDAS framework.

Slide 8

Application Scenario MIDAS is made out of countless, passages, center points, and PDAs that are associated remotely. Numerous occurrences of MIDAS work in security basic or managed situations, in which a given occasion must be conveyed to its beneficiaries under all conditions. The product framework ought to bolster autonomic come up short over.

Slide 9

Application Scenario One way to deal with giving come up short over backing is through the replication of programming segments, and consequently the administrations they give. Nonetheless, part replication in MIDAS is troublesome for the most part because of its asset obliged nature (i.e., restricted CPU, memory, battery, system transfer speed, and so on.). The current methodologies for giving adaptation to internal failure on customary desktop stages are regularly wasteful in this area. On top of this, given limited processing assets, segments must be reproduced specifically so as to accomplish the greatest advantage. In the meantime, while replication can enhance adaptation to internal failure, different QoS properties can debase because of the overhead of executing extra segments and keeping reproductions synchronized. At long last, regularly the usage many-sided quality of cutting edge come up short over backing (i.e., replication, synchronization, and recuperation) and its coupling with the application rationale

Slide 10

Architectural Middleware

Slide 11

Architectural Middleware Prism-MW Is a middleware paltform that backings architecural abstrations by giving execution level modules to speaking to and controlling each building component. This empowers direct mapping between a design and its usage. The shaded classes constitute the middleware center, an insignificant subset of Prism-MW that empowers usage and execution of the structures in a solitary location space.

Slide 12

Architectural Middleware Core - an insignificant subset of Prism-MW that empowers usage and execution of the designs in a solitary location space. Block - speak to the distinctive design building hinders that are depict. Engineering - records the setup of its constituent segments, connectors, and ports, and gives offices to expansion and expulsion. Occasions - are use to catch interchanges Request Reply Ports - are the loci of connection Request Reply Weld - is a connection between two ports. Parts - perform calculation and keep up their own particular inward state. Connectors – are utilized to control the steering of occasions among the joined parts

Slide 13

Architectural Middleware Prism-MW configuration is proposed to be extensible while keeping the center unaltered. The center builds are subclassed through particular classes. Each of which has reference to various theoretical classes Each unique class can ave numerous executions

Slide 14

Architectural Middleware

Slide 15

Service Discovery Service disclosure in the implanted and pervasive situations is a testing issue. 1) We regularly don't have the foggiest idea about the area of an administration supplier at outline time. 2) An administration supplier may get to be distracted because of equipment, programming and system disappointments. 3) The area of an administration supplier may change at runtime.

Slide 16

Small Fragment of the Midas Architecture

Slide 17

Advantages of Service Discovery Topology-based steering Events are not labeled with an administration supplier; rather, they are directed taking into account topology and steering arrangements. Partition of concerns Application rationale is totally isolated from the administrations revelation convention and its subtleties Efficiency By utilizing connectors, we can minimize the quantity of required ports. Scallability Service query is perform in a various leveled way An administration query solicitation is done locally first and the if not discovered then universally.

Slide 18

Replication and Failover Facilities Two sorts of replications Active replication. Every copy of an administration supplier forms an administration demand occasion and rturns a reaction to the customer. Duplication identification is important for the customer to get one and only reaction. Aloof replication. One and only of the copies, chose as essential copy , forms a solicitation occasion and sends a reaction to the customer. Warm uninvolved replication, the rest of the reinforcement reproductions are synchronized intermittently with the essential imitation's state Cold detached replication, a reinforcement copy is stacked into memory and its state instated from a log just if current essential reproduction comes up short.

Slide 19

Replication and Failover Facilities Advantages Efficiency Each solicitation occasion is steered to the connector appended to a worldwide administration supplier for reinforcement. Every one of the imitations get the same grouping of solicitation occasion Extensibility A worldwide administration supplier can basic send its state occasionally to every one of its reinforcements. Adaptability Addition and evacuation of replication backing can without much of a stretch accomplished at the engineering level by expansion or expulsion of a connector

Slide 20

Replication and Deployment Analysis The framework segments have differing degrees of unwavering quality, and the replication of an exceedingly dependably segment gives less advantage than the replication of an inconsistent segment. Every part can work together with different segments in various framework use cases. The framework equipment can fall flat and the execution and accessibility of the remote system is erratic. Once a segment has been recognized as a possibility for replication, a hub must be chosen for the copy's organization, thinking about the computational assets required and those accessible at every host. The last component of our coordinated adaptation to non-critical failure system is the determination of segment reliabilities, which affect vigorously the replication and arrangement calculations, and at last the adequacy of the runtime administration disclosure and failover support in Prism-MW.

Slide 21

Related Work ArchJava is an expansion to Java that binds together programming design with usage, guaranteeing that the execution fits in with building requirements Orbix/E is a lightweight CORBA ORB enhanced fro installed applications. XMIDDLE is an information sharing middleware for versatile registering (encoded as XML) MobiPADS is intelligent middleware that backings both dynamic sending of enlarged administrations for portable processing and element adjustment for giving adaptable setup of assets

Slide 22

Conclusions The improvement of deficiency tolerant programming frameworks in pervasive situations is a testing assignment. The many-sided quality of giving propelled adaptation to internal failure offices, for example, segment replication, copy synchronization, and failover, frequently brings about unbending and unmanageable applications. This spoke to a novel way to deal with displaying and executing adaptation to non-critical failure offices at the design level. This methodology brings about a partition of use rationale from adaptation to internal failure rationale, which thusly helps the development, examination, and adjustment of the product framework. We have utilized these attributes to quantitatively